Weekly Moving-Average Channel Mean Reversion with Model Filtering

Summary

The article outlines a EURUSD mean-reversion system that uses weekly moving averages of highs and lows to form a channel. A move above the upper average is treated as an overbought condition and a short candidate; a move below the lower average is treated as oversold and a long candidate. A statistical model is intended to filter these signals. The described workflow exports historical OHLC data, creates a future close-to-open ratio target, compares six regression approaches, and selects linear regression after flexible models perform worse on the test sample than on training data. The model is then converted to ONNX for use in MetaTrader 5.

The evidence described includes train and test RMSE comparisons and a target distribution centered near one; the supplied excerpt omits much of the trading implementation and detailed performance results. The document mentions training data from 2011–2019 and a 2020–2026 test period, but later describes a half-and-half split and fitting the selected model on all available data, leaving the precise evaluation process unclear. Its conclusions are therefore not enough to establish out-of-sample trading profitability.

Key ideas

• Weekly moving averages of highs and lows define channel boundaries for contrarian entry signals.
• A regression model is used as a confirmation filter for the channel signals.
• The article compares six regressors and reports that training performance can differ sharply from test performance.
• It exports the selected linear model to ONNX for integration with MetaTrader 5.
• The supplied material omits detailed trading results and contains ambiguity about the data split and final model fit.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.